---
name: onboarding-orchestrator
description: Run the new-hire checklist across HR, IT, and the manager, chasing whoever's step is late. Use this skill when: Offer accepted — New hire added to the onboarding tracker.
---

# Run new-hire onboarding across every system

Goal: run the new-hire checklist across every system and chase whoever's step is stuck.

Connected tools this skill uses: notion, googlesheets, slack, gmail, googlecalendar.

## When to use

- Offer accepted — New hire added to the onboarding tracker.
- Or when the user asks directly to run the new-hire checklist across every system and chase whoever's step is stuck.

## Method

Run onboarding for {{new_hire}}, starting as {{role_type}}.

1. When {{new_hire}}'s offer is marked accepted, build their onboarding
   checklist from the {{role_type}} template: every step, owner, and due
   date.
2. Add {{new_hire}} to the onboarding-orchestrator tracker view with all
   steps set to not-started.
3. Each day until the start date, check every system directly (IT ticket
   status, payroll setup, facilities badge, manager tasks) and update the
   tracker with what's done, overdue, or blocked.
4. Before marking any step complete, confirm it against the source system
   directly — for example, the IT ticket is closed, not just assigned.
   Never mark a step done on assumption.
5. Remind the owner of any step overdue by {{reminder_threshold_days}} days;
   escalate to {{people_ops_lead}} if it is still open
   {{escalation_threshold_days}} days later.
6. The morning of the start date, post a day-one-readiness summary to
   {{delivery_channel}}: what's confirmed, what's still open.

Never mark a step complete without confirming it in its source system, and
never provision access or send an offer-adjacent email without the People
Ops lead's sign-off.

## Onboarding checklist template

Build the full new-hire checklist for a role type, with an owner and due date on every step.

## Method
1. Look up the {{role_type}} template stored in memory.
2. List every step across HR, IT, facilities, and the manager for this role type.
3. Assign an owner to each step from the template's owner map.
4. Set a due date for each step, counted back from the start date.
5. If no template exists for this role type, draft one from the closest existing template and flag it for review.
6. Save the built checklist to the onboarding-orchestrator tracker view with every step set to not-started.

## Rules
- Every step needs one named owner — never leave a step unowned.
- Store and reuse role-type templates in memory (set on first run, then stored in memory); update them as the process changes.
- Flag a newly drafted template for human review before treating it as standard.
- Never skip a step because a past hire's checklist skipped it — check the current template each time.

## Output
A checklist added to the tracker view: one row per step, with owner, due date, and status set to not-started.

## Owner chasing policy

Remind a stuck owner on schedule, then escalate — without nagging early or letting a step go silent.

## Method
1. Check each open step's status daily against its source system, not the tracker alone.
2. Compare each step's age against {{reminder_threshold_days}}.
3. Send the owner a reminder once a step passes that threshold.
4. Track whether the reminder resolved the step within a further {{escalation_threshold_days}} days.
5. If still open, escalate to {{people_ops_lead}} with the step's history attached.
6. Update the tracker with every reminder and escalation sent.

## Rules
- Default reminder threshold is {{reminder_threshold_days}} days, escalation threshold is {{escalation_threshold_days}} days (set on first run, then stored in memory).
- Verify a step's status against its source system before reminding or escalating — never chase a step that is actually done.
- Escalate to {{people_ops_lead}}, not a broad channel, so the owner isn't publicly called out.
- Never send more than one reminder before escalating — a second silence goes to the lead.

## Output
A log entry per reminder and escalation sent, plus an updated tracker row reflecting each step's current status.

## Readiness reporting

Roll the full checklist into one clear readiness call the manager can act on the morning of day one.

## Method
1. Pull every step's current status from the tracker view.
2. Separate steps into confirmed-complete, still-open, and blocked.
3. Weigh which open steps are day-one-critical versus steps that can land in week one.
4. Write a short summary: overall status, then each open item with its owner.
5. Post the summary to {{delivery_channel}} the morning of the start date.
6. Flag any day-one-critical step still open as an explicit risk, not a footnote.

## Rules
- Call a new hire "ready" only when every day-one-critical step is confirmed-complete, never on assumption.
- Separate day-one-critical items from week-one items so the manager isn't chasing the wrong thing.
- Post the summary once, the morning of the start date — not a stream of partial updates.
- Keep the summary to what the manager needs to act on, not the full checklist detail.

## Output
A day-one-readiness summary: overall status line, then each open or blocked step with its owner, posted to {{delivery_channel}}.

## Recurring jobs

This skill pairs with scheduled jobs. Offer to set each one up; only create the ones the user approves:

- Onboarding checklist chase (Daily, while any new hire's onboarding is still open) — https://eluu.ai/cookbook/onboarding-orchestrator/jobs-setup/onboarding-checklist-chase.md